High-Performance Simulation Tools for Hemodynamics

• Applicant: Professor Dr. Martin Kronbichler
• Subject Area: Fluid Mechanics ; Mathematics
• Term: since 2023
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 524455704

Today, cardiovascular diseases are the cause of around a third of all human deaths according to the World Health Organisation. The treatment of many diseases can be aided by numerical simulation of blood flows in the context of computational fluid dynamics (CFD), combining insights from computational biomechanics and mathematical modeling to derive new insights into the complex processes in the human body. Despite intense progress in recent years, several CFD challenges are still unresolved, in particular the simulation turnaround times, the lack of accuracy due to over-simplification and poor robustness. These issues hold back the use of numerical tools to digitally investigate surgery variants, perform parameter studies on virtual cohorts, analyze otherwise inaccessible data, design medical devices, and much more. This proposal aims to overcome these barriers by novel CFD method development, integrating new hemodynamic-specific algorithms to the open-source, high-performance, matrix-free DG framework ExaDG. More precisely, new methods for simulating the incompressible flow of non-Newtonian fluids with discontinuous Galerkin methods, the simulation of hyperelastic materials representing arterial tissue and the fluid-structure interaction with accelerated partitioned methods will be devised. These developments will incorporate the major modeling aspects for real-world hemodynamic applications with a strong focus on computational efficiency. The main pillars of the new developments are high-order discretization methods, which enable a high solution accuracy already for coarser meshes due to their better dispersion behavior, thus reducing the communication bottlenecks in modern solvers in favor of additional local computations. A central ingredient is the development of matrix-free solver technologies, tailored to the properties of modern computer hardware including large-scale parallelism, which allow for a similar computational cost per unknown as established low-order approaches. The developed software will be made publicly available to further foster innovations and exchange of ideas within the community and to improve accessibility for the fluid dynamics, biomechanics, bio-engineering and medical sectors.

• DFG Programme: Research Grants
• International Connection: Austria
• Cooperation Partners: Professor Dr.-Ing. Thomas-Peter Fries; Professor Dr.-Ing. Gerhard A. Holzapfel